Email Delivery

Receive new posts as email.

Email address

Syndicate this site

RSS | Atom

Contact

About This Site
Contact Us
Privacy Policy

Search


November 2010
Sun Mon Tues Wed Thurs Fri Sat
  1 2 3 4 5 6
7 8 9 10 11 12 13
14 15 16 17 18 19 20
21 22 23 24 25 26 27
28 29 30        

Stories by Category

Basics :: Basics
Casting :: Casting Listen In Podcasts Videocasts
Culture :: Culture Hacking
Deals :: Deals
FAQ :: FAQ
Future :: Future
Hardware :: Hardware Adapters Appliances Chips Consumer Electronics Gaming Home Entertainment Music Photography Video Gadgets Mesh Monitoring and Testing PDAs Phones Smartphones
Industry :: Industry Conferences Financial Free Health Legal Research Vendor analysis
International :: International
Media :: Media Locally cached Streaming
Metro-Scale Networks :: Metro-Scale Networks Community Networking Municipal
Network Types :: Network Types Broadband Wireless Cellular 2.5G and 3G 4G Power Line Satellite
News :: News Mainstream Media
Politics :: Politics Regulation Sock Puppets
Schedules :: Schedules
Security :: Security 802.1X
Site Specific :: Site Specific Administrative Detail April Fool's Blogging Book review Cluelessness Guest Commentary History Humor Self-Promotion Unique Wee-Fi Who's Hot Today?
Software :: Software Open Source
Spectrum :: Spectrum 60 GHz
Standards :: Standards 802.11a 802.11ac 802.11ad 802.11e 802.11g 802.11n 802.20 Bluetooth MIMO UWB WiGig WiMAX ZigBee
Transportation and Lodging :: Transportation and Lodging Air Travel Aquatic Commuting Hotels Rails
Unclassified :: Unclassified
Vertical Markets :: Vertical Markets Academia Enterprise WLAN Switches Home Hot Spot Aggregators Hot Spot Advertising Road Warrior Roaming Libraries Location Medical Public Safety Residential Rural SOHO Small-Medium Sized Business Universities Utilities wISP
Voice :: Voice

Archives

November 2010 | October 2010 | September 2010 | August 2010 | July 2010 | June 2010 | May 2010 | April 2010 | March 2010 | February 2010 | January 2010 | December 2009 | November 2009 | October 2009 | September 2009 | August 2009 | July 2009 | June 2009 | May 2009 | April 2009 | March 2009 | February 2009 | January 2009 | December 2008 | November 2008 | October 2008 | September 2008 | August 2008 | July 2008 | June 2008 | May 2008 | April 2008 | March 2008 | February 2008 | January 2008 | December 2007 | November 2007 | October 2007 | September 2007 | August 2007 | July 2007 | June 2007 | May 2007 | April 2007 | March 2007 | February 2007 | January 2007 | December 2006 | November 2006 | October 2006 | September 2006 | August 2006 | July 2006 | June 2006 | May 2006 | April 2006 | March 2006 | February 2006 | January 2006 | December 2005 | November 2005 | October 2005 | September 2005 | August 2005 | July 2005 | June 2005 | May 2005 | April 2005 | March 2005 | February 2005 | January 2005 | December 2004 | November 2004 | October 2004 | September 2004 | August 2004 | July 2004 | June 2004 | May 2004 | April 2004 | March 2004 | February 2004 | January 2004 | December 2003 | November 2003 | October 2003 | September 2003 | August 2003 | July 2003 | June 2003 | May 2003 | April 2003 | March 2003 | February 2003 | January 2003 | December 2002 | November 2002 | October 2002 | September 2002 | August 2002 | July 2002 | June 2002 | May 2002 | April 2002 | March 2002 | February 2002 | January 2002 | December 2001 | November 2001 | October 2001 | September 2001 | August 2001 | July 2001 | June 2001 | May 2001 | April 2001 |

Recent Entries

In-Flight Wi-Fi and In-Flight Bombs
Can WPA Protect against Firesheep on Same Network?
Southwest Sets In-Flight Wi-Fi at $5
Eye-Fi Adds a View for Web Access
Firesheep Makes Sidejacking Easy
Wi-Fi Direct Certification Starts
Decaf on the Starbucks Digital Network
Google Did Snag Passwords
WiMax and LTE Not Technically 4G by ITU Standards
AT&T Wi-Fi Connections Keep High Growth with Free Service

Site Philosophy

This site operates as an independent editorial operation. Advertising, sponsorships, and other non-editorial materials represent the opinions and messages of their respective origins, and not of the site operator. Part of the FM Tech advertising network.

Copyright

Entire site and all contents except otherwise noted © Copyright 2001-2010 by Glenn Fleishman. Some images ©2006 Jupiterimages Corporation. All rights reserved. Please contact us for reprint rights. Linking is, of course, free and encouraged.

Powered by
Movable Type

« EarthLink Gains St. Petersburg (or Vice Versa) | Main | AirPort Extreme: Apple Breaks 90 Mbps »

February 16, 2007

How Draft N Makes Nice with Neighbors; 5 GHz Averts Tragedy of the Commons

The soon-to-be-approved Draft 2.0 for 802.11n will protect 2.4 GHz legacy networks well, but is tuned for 5 GHz: A large part of the horsetrading involved in the IEEE Task Group N's work between Jan. 2006 and Jan. 2007 centered on making sure that Draft N didn't beat up its older relatives, 802.11a, b, and g. The constraints placed on 2.4 GHz will naturally steer new network deployments into 5 GHz, a better band for video streaming around the home as well.

I spoke with Atheros, Broadcom, and Metalink recently, along with the Wi-Fi Alliance, to look at how the current Draft N will protect legacy networks, and where N will take the consumer markets. (Last month, I wrote lengthily about some of the technical issues that make 5 GHz appealing, but may also constrain it.)

What's Upgradable

The industry as a whole hasn't been happy with me for much of 2006. That's ok. Writing this site isn't a popularity contest, and to the credit of the many product managers and PR folks, they've always been willing to engage me politely on the points.

But it's a relief to say, at last, that I'm confident that the Draft N gear sold to date using chips from major manufacturers can be fully upgraded to Draft 2.0 via software and firmware patches. I'm finally hearing the kind of full endorsement that I wanted consumers to have before they purchased early 802.11n gear that had a chance of being incompatible at the highest possible speeds with a final standard.

"The changes that were made are all changes that can be accommodated with software driver changes," said Bill McFarland, chief technical officer at Atheros. "All of our products that we've ever shipped since Day One will be upgradable to the final draft 2.0," said Bill Bunch, Broadcom's director of product management for wireless LAN.

Wha's not being said at the moment, of course, is whether the current generation of equipment and that shortly to appear will also be upgradable to the final version of 802.11n. However, I believe bookmakers would take a bet against final upgradability--there will be too much at stake, even more than in this last period, to make that error. The industry prediction is that the path from Draft 2.0 to a final, ratified draft will be long--Oct. 2008 is the projected ratification date--but that few changes, none significant, are likely.

Those who own current generation Draft N gear, or are about to buy it, should know that there's no known timetable for the release of firmware updates that improve interoperability among devices from different makers, fix known glitches, and make changes to support elements of Draft 2.0 that aren't in current devices.

Broadcom and Atheros both noted that while they can have firmware updates completed in the near future, their customers--the device makers--have to integrate and release those upgrades in their own due time. Atheros's McFarland said, "I see no reason why, at least physically speaking, it couldn't be done within two months" on Atheros's end. But, he said, some makers may choose to push out a simple patch, while others integrate it with driver and host software releases. Tim Higgins of SmallNetBuilder spoke to major equipment manufacturers and compiled their responses.

Looking Backwards: Legacy Co-Existance

The current draft of 802.11n supports three separate mechanisms to keep Draft N devices from stamping all over legacy 802.11a, b, and g networks through its use of wide channels. The older standards support 20 MHz channels, and can carry a maximum of 54 Mbps of raw data per channel. 802.11n ups that to 65 Mbps, but, with at least two discrete data streams and double wide channels--40 MHz wide--has a raw rate of 300 Mbps. (I don't know how 65 times 4 equals 300, either, folks.)

Now 2.4 GHz already has a lot going on in it: 802.11b and g, of course; Bluetooth; microwave ovens; industrial, medical, and scientific (ISM) stuff; licensed ham radio in the lower part; other licensed purposes in other parts. Well, it's rather crowded in there.  (2.4 GHz has about 83.5 MHz in the US spread over 11 staggered channels, where we have 555 MHz available in the 5 GHz band spread over 23 distinct channels.)

It could be argued that 802.11n looks just like two b or g channels. But that's slightly specious because it's not likely that in your immediate area, you would suddenly see every neighbor add two access points at the same time, which would be the effect of 802.11n rolling out wide channels by default in 2.4 GHz. "We're not talking about whether 2.4 GHz will increase density, but what's the rate and what's the consequence when you achieve that limit," said Broadcom's Bunch.

As a result, some equipment makers are choosing to not allow wide channels in 2.4 GHz, Apple foremost among them. Apple explained this in a briefing with me a few weeks ago by noting that their wide support of Bluetooth was incompatible with using wide 802.11n channels, as Bluetooth--a frequency hopping standard that operates across the entire 2.4 GHz range--could see its throughput and even consistency affected.

Atheros's McFarland noted that that company's customers "obviously don't want to disrupt Bluetooth in some severe way" but they're willing to take some advantage of the wide channels, too. "40 MHz accelerates the problem of congestion and interference, but we already had the problem pretty good already," McFarland said. Broadcom's Bunch noted that the IEEE "took a fairly conservative approach wtih the algorithm while still allowing 40 MHz channels to be used," which lets equipment makers and customers choose what works in different circumstances.

For manufacturers who choose to deploy wide channels in 2.4 GHz, there will be three mechanisms that try to mitigate problems with legacy networks; one of these mechanisms won't be needed in 5 GHz, which is another factor in its favor.

Intolerance. We usually look on intolerance as a fault, but it's actually an advantage. The "intolerant bit" in Draft N allows any device on the same network to signal that it wants only 20 MHz channels to be used. That's how Apple will control its environemnt. "This device sending this packet would really like not to have 40 MHz going on around it in the 2.4 GHz band," said McFarland.

Clear-channel assessment (CCA). Every device that wants to transmit over a wide channel has to check that the additional 20 MHz of spectrum it might use isn't in use by a legacy network. McFarland noted that this mechanism is already in place for regular channels--"listen before talk"--and this extends it to the additional channel, allowing 20 MHz transmission if that part is clear.

Politeness. With politeness, there's a very low threshold that disables the use of 40 MHz channels when any legacy network is overlapping in the extended 20 MHz section. "If it's sending any amount of real data traffic at all" only regular channels can be used, McFarland said. Politeness extends for 30 minutes from the last check, and must be continuously monitored.

5 GHz Sensitivity Training: Rude, But Tolerant

There's no reason to be very polite, but neither to be intolerant in the 5 GHz band, chipmakers pointed out. The intolerance mechanism isn't required by the standard because there are enough channels to go around that there's no good reason for any given device to tell others to use regular channels.

And politeness goes out the window, too: the mechanism is still active, but it's less likely for those who use the default auto-channel selection option that will be found on perhaps all dual-band Draft N gateways that they will have traffic in the extra 20 MHz used in a wide channel. With 23 channels in play in the US, it's likely that a relatively empty segment should be available even in the heart of a city. "The probability of you being able to aquire a channel that is not being used is obviously significantly higher," said Ron Cates, vice president of marketing at Metalink, a firm specializes in media streaming using Draft N.

Those interviewed expect that dual-band gateways are likely to be much more prevalent than was the case with 802.11a by some orders of magnitude. Because 802.11a didn't hit the market until just before 802.11g shipped, there was no speed advantage to using 5 GHz, and dual-band a/b/g adapters weren't available immediately. Intel delayed its dual-band support for quite some time as the initial Centrino launch included just an 802.11b radio due to timing issues with the standard.

Intel wasn't caught flatfooted this time, and both Apple and Intel announced their general support for 802.11n within a week: Apple already shipping adapters with the N part disabled until they were secure in the standard being stable; Intel committing to dual-band a/b/g/n in their Centrino platform, with the first models due later this month.

Broadcom's Bunch noted about 5 GHz, "It does point the way to dual band products being more available than they have been. The basic issue you had since the beginning is that there's almost always a price premium, particularly on the infrastructure. On the notebook side, the price premium between a single band and a dual band has been quite low for some time." Bunch said that all PC makers that ship Broadcom Draft N adapters include dual-band N, which makes adoption of 5 GHz much simpler.

Ultimately, the desire to stream video will make 5 GHz dominant for home media adapters, just as Apple has signaled by including both 2.4 GHz and 5 GHz support in its Apple TV's Draft N adapter. Cates of Metalink noted that Internet Protocol Television (IPTV), the way in which providers are bringing TV to the home through fiber optic and other pipes, requires either 20 Mbps for older MPEG2 compression or 10 Mbps for newer MPEG4 compression to deliver full high-definition quality. That just won't fly in 2.4 GHz. The limitations of 2.4 GHz "compel us time and time again to tell our customers to configure" using 5 GHz if it's video they want to use, said Cates.

Choppers! (M*A*S*H Joke)

One small problem in 5 GHz appeared with Apple's initial release of the revised AirPort Extreme Base Station. While the gateway and the adapters that are built into Macs support 5 GHz, they only allow the use of eight of the 23 channels: the lowest four and the highest four. The reason? Apple won't state this, but it's because the middle 15 require the use of two protocols for broadcasting at the lowest necessary power level coupled with detecting and avoiding broadcasts on frequencies used by active military radar installations. That avoidance was part of a grand compromise in the US that opened 255 MHz of additional 5 GHz bandwidth.

McFarland noted that Atheros has been producing silicon that conforms to these standards for some time; it's the earliest firm that committed to 802.11a back when there were doubts that 802.11a could even be built using the least expensive chipmaking process, CMOS. He said, without specifically commenting on Apple's choice or plans, that adding radar avoidance and dynamic power control is as simple as a firmware upgrade for any modern Wi-Fi chipset.

Cates of Metalink said that, "If you're detecting the presence of signals there, you want to avoid those channels. That's the beauty of the 5 GHz band -- there's so many to choose from." Cates said that the radar in question in the US tends to be fixed and broadcasting continuously, making it simple to avoid. In Europe and Asia, Cates said, there's more use of bands also employed by civilians, making it perhaps more of an issue there. Cates said future mechanisms would allow gateways to signal clients to change channels, which would obviate increased density of both military and unlicensed 5 GHz use.

Draft N Certification and Beyond

Frank Hanzlik, executive director of the Wi-Fi Alliance, said that the group is well on track to deliver its first announcements of certified Draft N products by June, which tracks their goal for this program when announced last year.

In terms of the particulars of what will be tested, that's still up in the air. "We don't have an approved test plan yet; we're still in that process," Hanzlik said, although the group will clearly be looking at legacy and Draft N interaction, and devices' conformance with all the politeness and avoidance behaviors that are in the draft.

Beyond certification comes the emergence of faster Draft N devices that deploy more data streams. The first generation of gear can generate two simultaneous data streams that are directed by two or more transmitting antennas to occupy unique physical paths--so-called spatial multiplexing. The spec requires at least two streams, but can support up to four. The predictions are that those additional streams come slowly, with no near-term availability as the cost structure and demand isn't in place.

And, finally, comes ratification, now targeted for Oct. 2008. At which point, another, faster standard will likely already be under discussion and the subject of much debate.

4 Comments

I a little bit confused regarding 802.11n. All people are so exciting about this technology and every vendor start to create and launch 802.11n products.
I already heard about this since 2004 (when 802.11g was newly certificated). And I heard that we have to use 2 antennas (1 for transmitting and another for receiving).

Learn from the past, the speed of WLAN G (802.11g) is supposed to reach 54 mbps in normal mode and 104 mbps in turbo mode. But when we implement it, the speed only reach 24 mbps at normal mode and 56 mbps in turbo mode.

If this case also happens in WLAN N, then the credibility of WLAN will fall apart, and soon or later, no one can believe in WLAN speed.

I think WLAN N Drafters have to concern about this. What do you think?

Interesting questions, which show a good grasp of history. 802.11n uses at least four antennas: two each for sending and receiving. It can use more. That's not really a problem, as the antenna arrays don't seem to add considerable expense.

In this article, as with others I've written, I've made a clear distinction between over-the-air symbol rate--the raw number of bits that are transferred by the adapters--and the net throughput rate, or real data sent and received by computers on either end. (That's really application layer versus physical layer data rates, I suppose.)

With 802.11g, 54 Mbps was the raw symbol rate, and the only surprise was that throughput wasn't over 30 Mbps, but so much lower. Turbo mode and similar modes were never part of any IEEE standard, so any data rates there were only supported by individual vendors and their proprietary schemes.

The 300 Mbps rate for 802.11n, which is sometimes called a 240 Mbps for reasons I'm not sure of, is really about 100 Mbps of throughput in ideal cases. We may see more speed yet, but in the same realm in which a G device hits 20+ Mbps of throughput, N will hit 70 to 100 Mbps. I've seen this in testing.

Interesting article, but I'll dispute one point the vendors made. Yes, the chipset makers may very well be able to provide firmware and the like for upgrades to existing DraftN hardware. However, I have several wi-fi devices going back to early 11b for which similar claims were made. The problem is and always has been that the vendors profess a device is future proof. But, experience has shown that while technically a particular device has the capability the vendor has little incentive to bring forward new and improved features and operability. So, for the chipset manufacturers to say their chips can be upgraded it is a hollow promise. The words "can" and "will" are wholly separate. I'd much rather wait for certified hardware.

[Editor's note: I don't disagree with your desire for certainty--and certification. I would argue, though, that the stakes are different. I recall when WPA shipped, that it would be possible for devices shipped in 1999 and later with 802.11b to be upgraded to WPA. The TKIP key for 802.11i was designed with that in mind. However, some makers were out of business, others had sold their chip businesses, and others chose not to. The situation today is that companies have sold gear with the expectation among users that it will be upgradable to some extent, regardless of disclaimers on the hardware box. I think any hardware maker or chipmaker that failed to achieve upgrades would face a huge amount of negative publicity, and reduced marketshare.--gf]

I do standardization work in TGn. To clear your doubts on the throughput numbers:

With 800ns guard interval (GI) between symbols:

1 spatial stream (SS) in 20 MHz gives 65 Mbps.

2 SS - 20 MHz = 130.

1 SS in 40 Mhz gives 135.
2 SS in 40 Mhz gives 270.

With the optional 400ns GI:

1 spatial stream (SS) in 20 MHz gives 72 Mbps.
2 SS - 20 MHz = 144.
1 SS in 40 Mhz gives 150.
2 SS in 40 Mhz gives 300.

Hope this helps.